JP2019160030A - Plant control device - Google Patents

Abstract

To suppress deterioration of responsiveness of a control output while suppressing overshooting of the control output when the control output of a plant is brought close to a target value.SOLUTION: A plant control device includes: a target value calculation unit 2 that, based on a predetermined parameter d of a plant 6, calculates a target value r of a control output x of the plant; a target value correction unit 3 corrects the target value in a direction to suppress overshooting of the control output and calculates a corrected target value w; and a feedback controller 5 that determines a control input of the plant so that the control output approaches the corrected target value. When the control output is changed to the target value, the target value correction unit sets the corrected target value so that a correction amount of the target value is equal to or less than a predetermined value, and then changes the correction amount so that the correction amount becomes larger than the predetermined value, and then changes the corrected target value so that the correction amount of the target value becomes the predetermined value or less before the control output reaches the target value.SELECTED DRAWING: Figure 3

Description

  The present invention relates to a plant control apparatus.

  In the plant to be controlled, feedback control is performed so that the control output approaches the target value. However, in actual control, the value of the control output is often limited due to hardware or control restrictions. If such a restriction is ignored and the control system is designed, transient response may be deteriorated and control may become unstable.

  A reference governor is known as a technique for improving constraint satisfaction (for example, Patent Document 1). The reference governor calculates the corrected target value by correcting the target value of the control output calculated based on a predetermined parameter of the plant in consideration of constraint satisfaction. Specifically, the reference governor calculates a corrected target value by performing a minimum value search for a predetermined objective function.

  Patent Document 1 describes that a target value for a supercharging pressure and an EGR rate in a diesel engine is corrected by a reference governor. Specifically, the corrected target value is calculated by updating the corrected target value by the gradient method so that the value of the objective function becomes small.

  The objective function includes a term relating to the correction amount of the target value and a term relating to the degree of satisfaction of the constraint condition regarding the state quantity of the plant. The term relating to the degree of satisfaction of the constraint condition increases as the degree of satisfaction of the constraint condition decreases. When the term relating to the degree of satisfaction of the constraint condition is large, the target value is corrected so that the value of the objective function becomes small. As a result, the degree of satisfaction of the constraint condition is increased and the overshoot of the control output is also suppressed.

JP 2017-101627 A

  By the way, when control is performed to bring the control output closer to the target value when the difference between the target value of the control output and the current value is large, the state quantity of the plant is smaller than when the difference between the target value and the current value is small. The possibility that the predicted value in the future conflicts with the constraint condition increases. For this reason, if the minimum value search of the objective function is performed when the difference between the target value and the current value is large, the correction amount of the target value tends to increase. However, if the correction amount of the target value is increased when the difference between the target value and the current value is large, the speed at which the control output approaches the target value becomes slow, and the response of the control output deteriorates.

  Then, in view of the said subject, when the control output of a plant is brought close to a target value, it is in suppressing the deterioration of the responsiveness of a control output, suppressing the overshoot of a control output.

  The gist of the present disclosure is as follows.

  (1) A target value calculation unit that calculates a target value of the control output of the plant based on a predetermined parameter of the plant, and corrects the target value by correcting the target value in a direction that suppresses overshoot of the control output. A target value correcting unit for calculating, and a feedback controller for determining a control input of the plant so that the control output approaches the corrected target value, and the target value correcting unit changes the control output to the target value. The correction target value is set so that the correction amount of the target value is not more than a predetermined value, and then the correction target value is changed so that the correction amount of the target value is larger than the predetermined value. Then, before the control output reaches the target value, the plant control device changes the correction target value so that the correction amount of the target value is equal to or less than the predetermined value.

  (2) When changing the control output to the target value, the target value correcting unit corrects the target value when a difference between the target value and the current value of the control output is larger than a first reference value. The correction target value is set so that the amount becomes equal to or less than the predetermined value, and the correction value of the target value becomes larger than the predetermined value when the difference reaches the first reference value. When the target value is changed and the difference reaches the second reference value, the correction target value is changed so that the correction amount of the target value is equal to or less than the predetermined value, and the second reference value is The plant control device according to (1), which is smaller than one reference value.

  (3) The target value correction unit calculates the correction target value by performing a minimum value search of an objective function, and the objective function includes a term relating to a correction amount of the target value, and a state quantity of the plant. And when the difference between the target value and the current value of the control output is relatively large, the correction of the target value is performed as compared with the case where the difference is relatively small. The plant control apparatus according to (1), configured to increase a degree of contribution of the term relating to the amount to the value of the objective function.

  (4) The plant according to (3), wherein the term related to the correction amount of the target value is a value obtained by multiplying a component that increases as the correction amount of the target value increases by a component that increases as the difference increases. Control device.

  (5) The plant according to (3), wherein the term relating to the degree of satisfaction of the constraint condition is a value obtained by dividing a component that increases as the degree of satisfaction of the constraint condition decreases by a component that increases as the difference increases. Control device.

  (6) The plant control apparatus according to (4) or (5), wherein the component that increases as the difference increases is the nth power of the difference, and n is greater than zero.

  (7) The plant control device according to (6), wherein the plant is an internal combustion engine, and n is a value of 4 or more and 6 or less.

  ADVANTAGE OF THE INVENTION According to this invention, when the control output of a plant is brought close to a target value, the deterioration of the responsiveness of a control output can be suppressed, suppressing the overshoot of a control output.

FIG. 1 is a diagram illustrating a target value tracking control structure of the plant control apparatus according to the first embodiment. FIG. 2 shows a feedforward control structure obtained by equivalently modifying the target value tracking control structure of FIG. FIG. 3 is a time chart of the control output target value, the corrected target value, and the actual value when the control output is changed to the target value. FIG. 4 is a flowchart showing a control routine for target value correction processing in the first embodiment. FIG. 5 is a diagram illustrating a target value tracking control structure of the plant control apparatus according to the second embodiment. FIG. 6 is a time chart of the corrected target value when the value of n in the objective function is changed. FIG. 7 is a diagram schematically showing the degree of effect when the value of n in the objective function is changed when the plant is an internal combustion engine. FIG. 8 is a time chart of the control output target value, the corrected target value, and the actual value when the control output is changed to the target value. FIG. 9 is a flowchart showing a control routine for target value correction processing in the second embodiment.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the following description, the same reference numerals are assigned to similar components.

<First embodiment>
First, a first embodiment of the present invention will be described with reference to FIGS.

<Configuration of plant control device>
FIG. 1 is a diagram illustrating a target value tracking control structure of the plant control apparatus according to the first embodiment. The plant control apparatus includes a target value calculation unit 2, a target value correction unit 3, a comparison unit 4, and a feedback controller 5. For example, a microcomputer such as an electronic control unit (ECU) functions as a plant control device.

  A portion surrounded by a broken line in FIG. 1 functions as a closed loop system 10 that controls the output of the plant 6 to be controlled. When the closed loop system 10 has been designed, the feedforward control structure of FIG. 2 can be obtained by equivalently deforming the target value tracking control structure of FIG.

  The target value calculation unit 2 calculates a target value r of the control output x of the plant 6 based on the exogenous input d, and outputs the target value r to the target value correction unit 3. The target value calculation unit 2 is configured as a target value map showing the relationship between the exogenous input d and the target value r, for example. The exogenous input d is a predetermined parameter of the plant 6.

  When the target value r is directly input to the closed loop system 10, feedback control is performed so that the control output x approaches the target value r. In this case, the control output x can be quickly brought close to the target value r, but the overshoot amount of the control output x is increased. For this reason, in this embodiment, the target value r is corrected by the target value correcting unit 3. The target value correcting unit 3 corrects the target value r in a direction to suppress the overshoot of the control output x and calculates a corrected target value w.

  The comparison unit 4 calculates a deviation e (= w−x) by subtracting the control output x from the corrected target value w, and inputs the deviation e to the feedback controller 5. The correction target value w is input to the comparison unit 4 by the target value correction unit 3, and the control output x is output from the plant 6 to which the control input u and the exogenous input d are input. The control output x is detected by a detector such as a sensor or estimated using a calculation formula or the like.

  The feedback controller 5 determines the control input u so that the control output x approaches the corrected target value w. That is, the feedback controller 5 determines the control input u so that the deviation e approaches zero. The feedback controller 5 uses known controls such as PI control and PID control. The feedback controller 5 inputs the control input u to the plant 6. Further, the control output x is input to the feedback controller 5 as state feedback. Note that the input of the control output x to the feedback controller 5 may be omitted. Further, the comparison unit 4 may be incorporated in the feedback controller 5.

<Correction of target value>
As described above, the target value correcting unit 3 calculates the corrected target value w by correcting the target value r in a direction to suppress the overshoot of the control output x. The direction in which the overshoot of the control output x is suppressed means a direction in which the corrected target value w is decreased when the control output x is increased toward the target value r, and the control output x is set to the target value r. When decreasing in the direction, it means a direction in which the correction target value w is increased.

  If the correction amount of the target value r is increased when the difference between the target value r of the control output x and the current value is large, the speed at which the control output x approaches the target value r becomes slow, and the response of the control output x is reduced. Getting worse. In order to suppress overshoot of the control output x, it is effective to increase the correction amount of the target value r when the control output x approaches the target value r. In order to converge the control output x to the target value r, it is necessary to reduce the correction amount of the target value r before the control output x reaches the target value r.

  For this reason, in this embodiment, when changing the control output x to the target value r, the target value correction unit 3 sets the correction target value w so that the correction amount of the target value r is equal to or less than a predetermined value, and then The correction target value w is changed so that the correction amount of the target value r becomes larger than the predetermined value, and then the correction amount of the target value r becomes equal to or less than the predetermined value before the control output x reaches the target value r. The correction target value w is changed as follows. The correction amount of the target value r is a difference between the target value r and the correction target value w.

  As a result, when the difference between the target value r of the control output x and the current value is relatively large, the change speed of the control output x can be increased, and the deterioration of the responsiveness of the control output x is suppressed. be able to. Further, when the difference between the target value r of the control output x and the current value is relatively small, the change speed of the control output x can be reduced, and the overshoot of the control output x can be suppressed. Therefore, in this embodiment, when the control output x is brought close to the target value r, it is possible to suppress the deterioration of the response of the control output x while suppressing the overshoot of the control output x.

  In the present embodiment, when the difference between the target value r of the control output x and the current value is larger than the first reference value, the correction target value w is set so that the correction amount of the target value r is not more than a predetermined value, When the difference between the target value r and the current value reaches the first reference value, the correction target value w is changed so that the correction amount of the target value r is larger than the predetermined value, and the target value r and the current value are changed. When the difference between and reaches the second reference value, the correction target value w is changed so that the correction amount of the target value r is equal to or less than the predetermined value. The second reference value is smaller than the first reference value.

  For example, when the difference between the target value r and the current value is larger than the first reference value, the target value correcting unit 3 sets the corrected target value w to the target value r, and the difference between the target value r and the current value is When the first reference value is reached, the corrected target value w is moved away from the target value r, and when the difference between the target value r and the current value reaches the second reference value, the corrected target value w is changed to the target value r. change.

<Description of control using time chart>
FIG. 3 is a time chart of the control output target value, the corrected target value, and the actual value when the control output is changed to the target value. In FIG. 3, the target value r of the control output x is indicated by a two-dot chain line, and the corrected target value w of the control output x is indicated by a one-dot chain line. Further, the actual value (current value) xe of the control output x when the control of the present embodiment is executed is indicated by a solid line, and the actual value (current value) of the control output x when the control of the comparative example is executed. Value) xp is indicated by a broken line.

  In the example of FIG. 3, the state of the plant 6 changes at time t1, and the target value r increases. The target value r is maintained at a substantially constant value after time t1. In this example, control for increasing the control output x toward the target value r is performed.

  In the control of the comparative example, the target value r is not corrected. Therefore, the control input u is determined by the feedback controller 5 so that the control output x approaches the target value r. As a result, after time t1, the actual value xp gradually increases and reaches the target value r at time t4. After time t4, an overshoot of the control output x occurs, and then the actual value xp converges to the target value r.

  On the other hand, in the control of the present embodiment, the correction target value w is calculated, and the control input u is determined by the feedback controller 5 so that the control output x approaches the correction target value w. From time t1 to time t2, the corrected target value w is equal to the target value r. That is, the target value r is not corrected.

  At time t2, the correction target value w is changed so that the difference between the target value r and the actual value xe reaches the first reference value R1, and the correction amount of the target value r is larger than a predetermined value. In this example, the correction target value w is reduced stepwise so as to suppress overshoot of the control output x. The corrected target value w is maintained at a constant value from time t2 to time t3.

  At time t3, the difference between the target value r and the actual value xe reaches the second reference value, and the corrected target value w is changed to the target value r. In this example, the corrected target value w is increased stepwise so that the control output x converges to the target value r. After time t3, the corrected target value w is maintained at the target value r. That is, the target value r is not corrected.

  Thereafter, the actual value xe reaches the target value r at time t5. After time t5, a slight overshoot of the control output x occurs, and then the actual value xe converges to the target value r.

  In the control of this embodiment, the increase in the control output x can be braked by reducing the correction target value w at time t2. For this reason, the overshoot amount of the control output x can be reduced as compared with the control of the comparative example. Further, since the corrected target value w is set to the target value r before the time t2, the control output x can be quickly brought close to the target value r as in the control of the comparative example. Therefore, in the control of the present embodiment, it is possible to suppress the deterioration of the responsiveness of the control output x while suppressing the overshoot of the control output x.

<Target value correction processing>
Hereinafter, with reference to the flowchart of FIG. 4, the control for correcting the target value r and calculating the corrected target value w will be described in detail. FIG. 4 is a flowchart showing a control routine for target value correction processing in the first embodiment. This control routine is repeatedly executed at predetermined execution intervals by the plant control device.

  First, in step S101, the target value calculation unit 2 acquires an exogenous input d. Next, in step S102, the target value calculation unit 2 calculates a target value r based on the exogenous input d.

  Next, in step S103, the target value correcting unit 3 determines that the absolute value of the value obtained by subtracting the current value PV of the control output x from the target value r, that is, the difference between the target value r and the current value PV is greater than the first reference value R1. It is determined whether or not it is larger. The first reference value R1 is determined in advance and is a value larger than zero. When it is determined that the difference between the target value r and the current value PV is larger than the first reference value R1, the present control routine proceeds to step S104.

  In step S104, the target value correcting unit 3 sets the corrected target value change flag to 1. The corrected target value change flag F is reset to zero when the feedback control of the control output x is completed. Next, in step S105, the target value correcting unit 3 sets the corrected target value w to the target value r in order to quickly change the control output x to the target value r. After step S105, this control routine ends.

  On the other hand, when it is determined in step S103 that the difference between the target value r and the current value PV is equal to or less than the first reference value R1, the present control routine proceeds to step S106. In step S106, the target value correcting unit 3 determines whether or not the corrected target value change flag F is 1. When it is determined that the correction target value change flag F is 1, the present control routine proceeds to step S107.

  In step S107, the target value correction unit 3 changes the correction target value w so that the correction amount of the target value r is larger than a predetermined value. The predetermined value is determined in advance so as to suppress overshoot of the control output x, and is a value larger than zero. The target value correction unit 3 changes the correction target value w in a step shape in a direction to suppress the overshoot of the control output x. Specifically, when increasing the control output x toward the target value r, the target value correction unit 3 makes the correction target value w smaller than the target value r and moves the control output x toward the target value r. In the case of decreasing, the corrected target value w is made larger than the target value r.

  Next, in step S108, the target value correcting unit 3 determines whether or not the difference between the target value r and the current value PV is equal to or less than the second reference value R2. The second reference value R2 is determined in advance and is larger than zero and smaller than the first reference value R1. When it is determined that the difference between the target value r and the current value PV is greater than the second reference value R2, the present control routine ends. In this case, the correction target value w is maintained at the value changed in step S107.

  On the other hand, when it is determined that the difference between the target value r and the current value PV is equal to or smaller than the second reference value R2, the present control routine proceeds to step S109. In step S109, the target value correcting unit 3 changes the corrected target value w to the target value r in order to converge the control output x to the target value r. The target value correction unit 3 changes the correction target value w to the target value r in steps. Next, in step S110, the target value correcting unit 3 sets the corrected target value change flag F to zero. After step S110, the control routine ends.

  If it is determined in step S106 that the correction target value change flag F is zero, the present control routine proceeds to step S111. In step S111, the target value correcting unit 3 sets the corrected target value w to the target value r. After step S111, this control routine ends.

  In step S105, step S109, and step S111, the correction target value w may be set to a value different from the target value r if the correction amount of the target value r is equal to or smaller than a predetermined value. Further, the target value correcting unit 3 may gradually change the corrected target value w to the changed value in step S107 and gradually change the corrected target value w to the target value r in step S109.

<Second embodiment>
The configuration and control of the plant control apparatus in the second embodiment are basically the same as those of the plant control apparatus in the first embodiment except for the points described below. For this reason, the second embodiment of the present invention will be described below with a focus on differences from the first embodiment.

  FIG. 5 is a diagram illustrating a target value tracking control structure of the plant control apparatus according to the second embodiment. In the second embodiment, the reference governor (RG) 30 functions as a target value correcting unit that corrects the target value r and calculates the corrected target value w. Y in FIG. 5 is a state quantity of the plant 6 in which a possible value is restricted.

  As described above, in the closed loop system 10, feedback control of the control output x is performed. However, in actual control, the state quantity y is limited due to hardware or control restrictions. For this reason, when the target value calculated without considering the constraint is input to the closed-loop system 10, the state quantity y conflicts with the constraint, which may cause deterioration of transient response and instability of control.

  For this reason, the reference governor 30 corrects the target value r so as to increase the degree of satisfaction of the constraint condition for the state quantity y, and calculates the corrected target value w. Specifically, the reference governor 30 calculates the corrected target value w by performing a minimum value search of the objective function. In the minimum value search, the reference governor 30 calculates the final correction target value w to be input to the closed loop system 10 by updating the correction target value w a predetermined number of times so that the value of the objective function becomes small.

  The objective function includes a term relating to the correction amount of the target value r and a term relating to the degree of satisfaction of the constraint condition for the state quantity y. When the difference between the target value r of the control output x and the current value is relatively large, in order to increase the degree of satisfaction of the constraint condition compared to when the difference between the target value r and the current value is relatively small, It is necessary to increase the correction amount of the target value r.

  However, if the correction amount of the target value r is increased when the difference between the target value r and the current value is large, the speed at which the control output x approaches the target value r becomes slow, and the responsiveness of the control output x deteriorates. . In order to suppress overshoot of the control output x, it is effective to increase the correction amount of the target value r when the control output x approaches the target value r.

  For this reason, the objective function indicates that the difference between the target value r of the control output x and the current value is relatively large compared to the case where the difference between the target value r of the control output x and the current value is relatively small. Thus, the correction amount of the target value r is configured to be small. Specifically, the objective function indicates that when the difference between the target value r and the current value of the control output x is relatively large, compared to when the difference between the target value r and the current value is relatively small, The contribution to the value of the objective function in the term relating to the correction amount of the target value is configured to be large.

  As a result, when the difference between the target value r of the control output x and the current value is relatively large, the change speed of the control output x can be increased, and the deterioration of the responsiveness of the control output x is suppressed. be able to. Further, when the difference between the target value r of the control output x and the current value is relatively small, the change speed of the control output x can be reduced, and the overshoot of the control output x can be suppressed. Therefore, in the second embodiment, when the control output x is brought close to the target value r, it is possible to suppress the deterioration of the response of the control output x while suppressing the overshoot of the control output x.

For example, the objective function J (w) is defined by the following formula (1).
J (w) = (r−w) ² (| r−PV |) ⁿ + S ₁ ² + S ₂ ² +... (1)
Here, PV is the current value of the control output r, and n is a value greater than zero. The objective function J (w) is a term relating to the correction amount of the target value r (the first term on the right side of the equation (1)) and a term relating to the degree of satisfaction of the constraint on the state quantity y (the second item on the right side of the equation (1)). It is composed as the sum of

The term relating to the correction amount of the target value r becomes a component ((r−w) ² ) that increases as the correction amount of the target value r increases, and increases as the difference between the target value r of the control output x and the current value increases. This is a value obtained by multiplying the component ((| r−PV |) ⁿ ). In this case, when the correction amount of the target value r is a predetermined value, the term relating to the correction amount of the target value r increases as the difference between the target value r of the control output x and the current value increases. Therefore, when the difference between the target value r of the control output x and the current value is relatively large, the target value is larger than when the difference between the target value r of the control output x and the current value is relatively small. The contribution to the value of the objective function J (w) in the term relating to the correction amount of r is increased.

  In the above equation (1), the component that increases as the correction amount of the target value r increases is the difference between the target value r and the correction target value w, that is, the square of the correction amount of the target value r. However, this component may be a value obtained by multiplying the correction amount of the target value r by a predetermined coefficient.

  In the above equation (1), the component that increases as the difference between the target value r of the control output x and the current value increases is the nth power of the difference between the target value r and the current value PV. However, this component may be a value obtained by multiplying the difference between the target value r and the current value PV by a predetermined coefficient.

The term relating to the degree of satisfaction of the constraint condition includes the square of the first penalty function S ₁ , the square of the second penalty function S ₂ , and the like. The number of penalty functions varies depending on the number of constraints. For example, when the number of constraint conditions is four, the terms related to the degree of satisfaction of the constraint conditions are the square of the first penalty function S ₁ , the square of the second penalty function S ₂ , the square of the third penalty function S ₃ and the second 4 includes a square of the penalty function S _4. Each penalty function becomes larger as the degree of satisfaction of the constraint is lower. Each penalty function may not be squared.

The first penalty function S ₁ relates to the degree of satisfaction of the constraint condition for the control output x that is one of the state quantities y. In the first penalty function S ₁ , it is defined as a constraint condition that the overshoot amount of the control output x does not become larger than zero. When the control output x is increased toward the target value r, the first penalty function S ₁ is defined by the following equation (2).

Here, x (k) is a future predicted value of the control output x, and p ₁ is a predetermined weight coefficient. K is a discrete time step, and Nh is the number of prediction steps (prediction horizon). The first penalty function S ₁ has a difference between the future predicted value x (k) of the control output x and the target value r when the future predicted value x (k) of the control output x is larger than the target value r. A penalty is added to the objective function J (w). For this reason, the first penalty function S ₁ increases as the overshoot amount of the future predicted value x (k) of the control output x increases.

On the other hand, when the control output x is decreased toward the target value r, the first penalty function S ₁ is defined by the following equation (3).

In this case, when the future predicted value x (k) of the control output x is smaller than the target value r, the first penalty function S ₁ is calculated using the future predicted value x (k) of the control output x and the target value r. Is added to the objective function J (w) as a penalty. For this reason, the first penalty function S ₁ increases as the overshoot amount of the future predicted value x (k) of the control output x increases.

The reference governor 30 calculates the future predicted value x (k) of the control output x using the model of the plant 6. For example, the reference governor 30 calculates the future predicted value x (k) of the control output x by the following equation (4).
x (k + 1) = f ₁ (x (k), w, d) (4)

Here, f ₁ is a model function used for calculating the future predicted value x (k) of the control output x. First, the predicted value x (1) of the control output x one step ahead from the calculation time is calculated using x (0) that is the control output x at the calculation time. The control output x (0) at the time of calculation is detected by a detector such as a sensor or estimated using a calculation formula or the like. Thereafter, the future predicted value x (k) of the control output x is sequentially calculated from the calculation time point to the predicted value x (Nh) of the control output x of Nh steps ahead, and the future predicted values of the total Nh control outputs x are calculated. The A value obtained by multiplying the time corresponding to one step by the number of prediction steps Nh is the prediction period.

The second penalty function S ₂ relates to the degree of satisfaction of the constraint condition for the predetermined state quantity y ₂ . The state quantity y ₂ is, for example, pressure, rotation speed, and the like. In the second penalty function S ₂ , for example, it is defined as a constraint condition that the state quantity y ₂ does not become larger than the upper limit value. In this case, the second penalty function S ₂ is defined by the following equation (5).

Here, y ₂ (k) is a future predicted value of the state quantity y ₂ , y _2up is a predetermined upper limit value of the state quantity y ₂ , and p ₂ is a predetermined weight coefficient. K is a discrete time step, and Nh is the number of prediction steps (prediction horizon). Second penalty function S _2, when the future predicted value y ₂ state quantity y ₂ where (k) is greater than the upper limit value y _2up, future state quantity y ₂ predicted value y ₂ (k) and the upper limit value The difference from _y2up is added to the objective function J (w) as a penalty. Therefore, the second penalty function S ₁ increases as the amount by which the future predicted value y ₂ (k) of the state quantity y ₂ exceeds the upper limit value y _2up increases.

ここで、ｙ_2lowは予め定められた状態量ｙ₂の下限値である。 In the second penalty function S ₂ , when it is defined as a constraint that the state quantity y ₂ is not smaller than the lower limit value, the second penalty function S ₂ is defined by the following equation (6).

Here, y _2low is a predetermined lower limit value of the state quantity y ₂ .

In this case, the second penalty function S _2, the state quantity y ₂ future predicted value y ₂ (k) if is smaller than the lower limit value y _2Low, future predicted value y ₂ of the state quantity y ₂ (k) And the lower limit y _2low are added as a penalty to the objective function J (w). For this reason, the second penalty function S ₁ increases as the amount by which the future predicted value y ₂ (k) of the state quantity y ₂ falls below the lower limit value y _2low increases.

The future predicted value y ₂ (k) of the state quantity y ₂ is calculated using the model of the plant 6 in the same manner as the future predicted value x (k) of the control output x. Further, the future predicted value of each state quantity may be calculated by other methods such as machine learning using a neural network.

  FIG. 6 is a time chart of the corrected target value when the value of n in the objective function J (w) is changed. In FIG. 6, the target value r is indicated by a broken line. The correction target value w when n is zero is indicated by a solid line, the correction target value w when n is 2 is indicated by a broken line, and the correction target value w when n is 4 is a one-dot chain line. The correction target value w when n is 6 is indicated by a solid line, and the correction target value w when n is 8 is indicated by a two-dot chain line. The correction target value w when n is zero is described as a comparative example.

  In the example of FIG. 6, the state of the plant 6 changes at time t1, and the target value r increases. The target value r is maintained at a substantially constant value after time t1. When n is zero, the correction amount of the target value r gradually decreases as the control output x approaches the target value r.

  On the other hand, when n is 2 to 8, if the difference between the target value r of the control output x and the current value is relatively large in the period from time t1 to time t2, the target of the control output x The correction amount of the target value r is smaller than when the difference between the value r and the current value is relatively small. This is related to the correction amount of the target value when the difference between the target value r of the control output x and the current value is relatively large compared to when the difference between the target value r and the current value is relatively small. This is because the objective function is configured so that the degree of contribution of the term to the value of the objective function is large. Further, after time t2, even if the target value r is not corrected, the amount of correction of the target value r gradually decreases as the term regarding the degree of satisfaction of the constraint condition decreases.

  When n is 2 to 8, the larger the number of n, the slower the timing for increasing the correction value of the target value and the faster the change speed of the correction target value w when increasing the correction value of the target value. . FIG. 7 is a diagram schematically showing the degree of effect when the value of n in the objective function J (w) is changed when the plant is an internal combustion engine. The effect increases as the overshoot amount of the control output x decreases, and increases as the response of the control output x improves. That is, the effect becomes the highest when the overshoot and response of the control output x are improved in the most balanced manner.

  As shown in FIG. 7, when n is 6 or more, the effect is saturated, and even if n is increased, the effect is hardly increased. Further, the calculation load of the reference governor 30 increases as n increases. For this reason, when the plant 6 is an internal combustion engine, it is preferable to set n to a value between 4 and 6. As a result, it is possible to effectively improve the overshoot and responsiveness of the control output x while reducing the calculation load of the reference governor 30.

<Description of control using time chart>
FIG. 8 is a time chart of the control output target value, the corrected target value, and the actual value when the control output is changed to the target value. In FIG. 8, the target value r of the control output x is indicated by a two-dot chain line, and the corrected target value w of the control output x is indicated by a one-dot chain line. Further, the actual value (current value) xe of the control output x when the control of the present embodiment is executed is indicated by a solid line, and the actual value (current value) of the control output x when the control of the comparative example is executed. Value) xp is indicated by a broken line.

  In the example of FIG. 8, the objective function of the above equation (1) is used to calculate the corrected target value w, and n is set to 7. In the example of FIG. 8, the state of the internal combustion engine changes at time t1, and the target value r increases. The target value r is maintained at a substantially constant value after time t1. In this example, control for increasing the control output x toward the target value r is performed.

  In the control of the comparative example, the target value r is not corrected. Therefore, the control input u is determined by the feedback controller 5 so that the control output x approaches the target value r. As a result, after time t1, the actual value xp gradually increases and reaches the target value r at time t4. After time t4, an overshoot of the control output x occurs, and then the actual value xp converges to the target value r.

  On the other hand, in the control of this embodiment, the correction target value w is calculated by the reference governor 30, and the control input u is determined by the feedback controller 5 so that the control output x approaches the correction target value w. As a result of calculating the correction target value w so as to reduce the value of the objective function, the correction target value w is gradually reduced from time t2 to time t3, and the correction amount of the target value r is gradually increased.

  Further, from time t3 to time t5, the correction target value w is gradually increased to the target value r, and the correction amount of the target value r is gradually decreased. After time t5, the corrected target value w is maintained at the target value r. The actual value xe gradually increases and reaches the target value r at time t5. In addition, after time t3, the actual value xe increases along the corrected target value w. After time t5, a slight overshoot of the control output x occurs, and then the actual value xe converges to the target value r.

  In the control of this embodiment, the increase in the control output x can be braked by gradually decreasing the correction target value w from time t2. For this reason, the overshoot amount of the control output x can be reduced as compared with the control of the comparative example. Further, since the correction amount of the target value r is small before the time t2, the control output x can be quickly brought close to the target value r as in the control of the comparative example. Therefore, in the control of the present embodiment, it is possible to suppress the deterioration of the responsiveness of the control output x while suppressing the overshoot of the control output x.

<Target value correction processing>
FIG. 9 is a flowchart showing a control routine for target value correction processing in the second embodiment. This control routine is repeatedly executed at predetermined execution intervals by the plant control device.

  First, in step S201, the target value calculation unit 2 acquires an exogenous input d. Next, in step S202, the target value calculation unit 2 calculates a target value r based on the exogenous input d.

Then, in step S203, the reference governor 30 sets the initial value w ₀ of the correction target value w to the target value r. Next, in steps S204 to S206, the reference governor 30 searches for the minimum value of the objective function.

  Specifically, in step S204, the reference governor 30 updates the corrected target value w so that the value of the objective function becomes small. For example, the reference governor 30 updates the correction target value w by the binary search method. The correction target value w may be updated by other known methods. For example, when the number of the control output x and the correction target value w is two or more, the reference governor 30 may update the correction target value w by the gradient method.

  Next, in step S205, the reference governor 30 adds 1 to the update count Count. The update count Count indicates the number of times the correction target value w has been updated in the minimum value search. The initial value of the update count Count is 0.

  Next, in step S206, the reference governor 30 determines whether or not the update count Count is equal to or greater than the predetermined count N. The predetermined count is, for example, 5 to 200. If it is determined in step S206 that the update count Count is less than the predetermined count, the control routine returns to step S204. Therefore, in the minimum value search, the correction target value w is repeatedly updated until the update count Count reaches the predetermined count N.

  If it is determined in step S206 that the update count Count is equal to or greater than the predetermined count N, the present control routine proceeds to step S207. In step S <b> 207, the reference governor 30 inputs the final correction target value w to the closed loop system 10. Next, in step S208, the reference governor 30 resets the update count Count to zero. After step S208, this control routine ends.

<Other embodiments>
The preferred embodiments according to the present invention have been described above, but the present invention is not limited to these embodiments, and various modifications and changes can be made within the scope of the claims.

For example, in the second embodiment, the objective function J (w) may be defined by the following equation (7).
J (w) = (r−w) ² + (S ₁ ² + S ₂ ² +...) / (| R−PV |) ⁿ (7)
Here, PV is the current value of the control output r, and n is a value greater than zero. The objective function J (w) is a term relating to the correction amount of the target value r (the first term on the right side of the equation (1)) and a term relating to the degree of satisfaction of the constraint on the state quantity y (the second item on the right side of the equation (1)). )).

  In this case, the term relating to the correction amount of the target value r is a component that increases as the correction amount of the target value r increases, and is the square of the correction amount of the target value r. Note that this component may be a value obtained by multiplying the correction amount of the target value r by a predetermined coefficient, or the like.

On the other hand, the term relating to the degree of satisfaction of the constraint condition is a component (S ₁ ² + S ₂ ² +...) That increases as the degree of satisfaction of the constraint condition decreases, and the difference between the target value r of the control output x and the current value. It is a value obtained by dividing by a component ((| r−PV |) ⁿ ) that increases as the value increases. In this case, when the difference between the target value r of the control output x and the current value is relatively large, the constraint condition is larger than when the difference between the target value r of the control output x and the current value is relatively small. The degree of contribution to the value of the objective function J (w) in the term relating to the degree of satisfaction of is small. In other words, when the difference between the target value r of the control output x and the current value is relatively large, the target value is larger than when the difference between the target value r of the control output x and the current value is relatively small. The contribution to the value of the objective function J (w) in the term relating to the correction amount of r is increased.

  In the above formula (7), the component that becomes larger as the degree of satisfaction of the constraint is lower is the sum of the squares of the penalty functions. However, this component may be the sum of the penalty functions.

  In the above equation (7), the component that increases as the difference between the target value r of the control output x and the current value increases is the nth power of the difference between the target value r and the current value PV. However, this component may be a value obtained by multiplying the difference between the target value r and the current value PV by a predetermined coefficient.

  Further, when the plant 6 is an internal combustion engine, the degree of effect when the value of n in the above equation (7) is changed is the same as the result shown in FIG. For this reason, it is preferable to set n to a value of 4 or more and 6 or less even when the objective function of the above equation (7) is used to calculate the correction target value w. As a result, it is possible to effectively improve the overshoot and responsiveness of the control output x while reducing the calculation load of the reference governor 30.

  Moreover, the plant control apparatus in this embodiment can be applied to any plant that can estimate the future predicted value of the state quantity of the control output. For example, the plant 6 may be a diesel engine and the control output x may be a supercharging pressure. In this case, the exogenous input d is the engine speed and the fuel injection amount, and the control input u is the opening of the variable nozzle provided in the turbine of the turbocharger.

  Further, the plant 6 may be a diesel engine, and the control output x may be a supercharging pressure and an EGR rate. In this case, the exogenous input d is the engine speed and the fuel injection amount, and the control input u is the opening of the variable nozzle, the opening of the throttle valve, and the opening of the EGR valve. The plant may be an internal combustion engine other than a diesel engine such as a gasoline engine, a vehicle, a machine tool, or the like.

2 Target Value Calculation Unit 3 Target Value Correction Unit 5 Feedback Controller 6 Plant

Claims (7)

A target value calculation unit for calculating a target value of the control output of the plant based on a predetermined parameter of the plant;
A target value correcting unit for correcting the target value in a direction to suppress overshoot of the control output and calculating a corrected target value;
A feedback controller for determining a control input of the plant so that the control output approaches the corrected target value;
When changing the control output to the target value, the target value correction unit sets the correction target value so that the correction amount of the target value is equal to or less than a predetermined value, and then the correction amount of the target value is The correction target value is changed so as to be larger than the predetermined value, and then the correction target value is set so that the correction amount of the target value becomes equal to or less than the predetermined value before the control output reaches the target value. Change the plant control device.   The target value correcting unit, when changing the control output to the target value, if the difference between the target value and the current value of the control output is larger than a first reference value, the correction amount of the target value is The correction target value is set to be equal to or less than a predetermined value, and when the difference reaches the first reference value, the correction target value is set so that the correction amount of the target value is larger than the predetermined value. When the difference reaches the second reference value, the correction target value is changed so that the correction amount of the target value is less than or equal to the predetermined value, and the second reference value is the first reference value The plant control device according to claim 1, which is smaller. The target value correction unit calculates the correction target value by performing a minimum value search of an objective function,
The objective function includes a term relating to the correction amount of the target value and a term relating to the degree of satisfaction of the constraint condition for the state quantity of the plant, and the difference between the target value and the current value of the control output is relatively 2. The plant according to claim 1, wherein when the difference is larger, the contribution of the term relating to the correction amount of the target value to the value of the objective function is larger than when the difference is relatively small. Control device.   The plant control device according to claim 3, wherein the term relating to the correction amount of the target value is a value obtained by multiplying a component that increases as the correction amount of the target value increases by a component that increases as the difference increases.   4. The plant control apparatus according to claim 3, wherein the term relating to the satisfaction degree of the constraint condition is a value obtained by dividing a component that increases as the satisfaction degree of the constraint condition decreases by a component that increases as the difference increases.   The plant control apparatus according to claim 4 or 5, wherein a component that increases as the difference increases is an nth power of the difference, and n is greater than zero.   The plant control device according to claim 6, wherein the plant is an internal combustion engine, and n is a value of 4 or more and 6 or less.

Images